Descriptors for Organometallic Complexes - Fundamental QSARs for Metal Ions - page 98

melting point, biochemical activity, or viscosity are not descriptors in this sense.
The statistical methodology quantitative structure-property relationship (QSPR)
allows mathematical formulae to be obtained for macroscopic properties based on
certain descriptors. In the QSPR abbreviation property (P) is often biochemical
activity (A). The vast majority of QSPR programs, calculations, and equations are
quantitative structure-activity relationship (QSAR) programs, calculations, and
equations.
Structure is the term applied here to specify the position in 3D space of the
atomic nuclei in the minimum energy conformer of the analyzed species. The
computer-assisted searching approach of the structure is called geometry optimiza-
tion . The correctness of geometry optimization is important because the value of
some descriptors depends on the structure.
4.2 METHODS AND COMPUTER PROGRAMS
Within the software that optimizes geometry and calculates descriptors, organo-
metallic complexes are treated as only a special type of organic species, that is,
a set of atomic nuclei incorporated in a common molecular electronic cloud like
“raisins in a cake”. The results of geometry optimization and descriptor calculations
are sufficiently accurate if these computer programs are properly parameterized for
all elements and chemical bonds in the organometallic compound of interest.
The first aim of the molecular/quantum mechanics programs (e.g., Personal
Computer MODELling, PCModel ® ; Molecular Orbital PACkage, MOPAC ® ;
HyperChem ® ) is geometry optimization. The minimum energy conformer is then
characterized by calculating a small number of descriptors characteristic of any par-
ticular program. Only descriptors that reflect the presence of the metal atom and
the coordinative bonds could be considered specific descriptors for organometallic
complexes.
Calculation of descriptors is an essential step in the operation of programs for
QSPR calculations. Such programs are diverse (Mekenyan et al. 1990; SciQSAR ® ,
Murugan et al. 1994; Cerius 2® ; TSAR ® ; Tarko 2005; Tarko 2008a) and include spe-
cialized modules for the calculation of a large number of descriptors. Some pro-
grams are used only for the calculation of descriptors (DRAGON ® ) that are later
used for QSPR calculations. It should be noted that most of the programs used for
calculation of descriptors and QSPR calculations are not parameterized for metal
atoms (M) and M-E chemical bonds.
The algorithm of the statistical software used for QSAR calculations selects
the significant descriptors by correlation with the biochemical activity and other
descriptors. They identify significant molecular fragments and select the QSAR
equations according to their quality. They also identify intramolecular synergetic
effects and assess quantitatively the representative sample feature of the training set.
They identify atypical molecules of what are called the outliers for lead hopping
category (Cramer et al. 2004; Saeh et al. 2005) and they estimate the value of the
biochemical activity for new, yet to be synthesized, molecules. Although relevant to
the topic of this chapter, presentation of QSAR methodology is not the subject of
this chapter.